Prepare the wording
AI can help shape supplied facts into a readable message. The person using it decides what information is appropriate to include and checks the meaning, commitments, and tone.
One of our own AI workflows is deliberately modest: help prepare email copy, put it into a consistently formatted Outlook draft, and leave the decision to send with a person.
By Battle Creek IT · Published · Updated
Writing help and email preparation are different jobs. An AI tool can help organize a reply, but the result still needs the correct recipient, subject, wording, and formatting. Pasting between tools can turn an otherwise ordinary email into a font repair exercise.
At BCIT, the repeatable part is a Microsoft Power Automate helper that accepts a recipient, subject, and plain-text body, then creates an HTML Outlook draft using our preferred Aptos 12-point formatting. It has no send action. The draft remains available for human review.
This is an example from our own internal workflow, not a client case study or a claim about measured time savings. The useful lesson is how the work is divided and where the person stays in control.
AI can help shape supplied facts into a readable message. The person using it decides what information is appropriate to include and checks the meaning, commitments, and tone.
The helper formats ordinary text into an Outlook draft. It escapes special characters for the HTML body and applies the chosen formatting consistently. This is a repeatable formatting task.
A person checks the saved draft and decides whether it should be sent. Creating a draft does not authorize sending it. There is no automatic send step in this helper.
Consider an illustrative message arranging a follow-up conversation. The person supplies the facts and asks for concise wording. AI prepares a draft, but should not invent an appointment time, promise a service outcome, or add a recipient from guesswork.
The approved wording then goes through the formatting helper. Before sending, the person checks the address, subject, actual saved body, and any attachments the message requires. If the wording creates a commitment the business did not approve, it gets corrected.
The example is intentionally ordinary. A useful first AI workflow does not need access to every business system or permission to act on its own. It needs a clear input, a bounded job, and an output someone can inspect.
A polished draft can still contain a wrong fact. Formatting cannot verify a customer’s request, decide whether a statement is appropriate, or establish that an attachment belongs in the message. Those decisions remain part of the review.
Existing replies also need care. A helper for creating a new draft should not be treated as proof that it will preserve a thread’s recipients, attachments, signature, or quoted history. The method needs to fit the particular email task.
Business information needs its own rules too. Choose approved tools and settings, limit the information used to what the task needs, and decide which records should stay outside the workflow. A draft-only design reduces one kind of risk; it does not replace data-handling decisions.
If a template solves the problem, use a template. If the output needs so much checking that it creates more work, change the approach. Our AI consulting and training starts with that kind of practical decision, alongside appropriate Microsoft 365 administration and business access controls.
No. The helper described here creates an HTML Outlook draft and has no send action. A person reviews the saved message and decides whether to send it.
No. Templates and ordinary automation are often enough for predictable messages and formatting. AI is useful only when it helps with the actual writing task and the required review remains practical.
Start with a free business conversation and initial onsite technology assessment with network discovery.
Request My Free AssessmentCall (269) 261-0269